OpenVMS System Management Guide, 2nd Edition

When you have a performance problem and you have identified the blocking resource, the next step is to consider load-balancing options. Load-balancing activities include the following:
Distributing load among systems in a VMScluster
Shifting high-priority interactive activities to lower-priority batch jobs
Delaying processing until off-peak periods (e.g., night or weekends)
Many computer systems sit idle from 5 P.M. to 9 A.M. This idle time represents more than 60 percent of the total potential capacity of a system. While you certainly cannot expect users to wait until midnight to edit an important document, you should carefully identify less critical system activities that can be performed during off hours as batch jobs. This will free scarce resources for users during normal business hours.
If you have multiple stand-alone systems or a VMScluster, you should ensure that user workloads are distributed as evenly as possible across available systems. You may have one system that is fully utilized and another system that is only 50 percent utilized. By distributing the load equally, you can increase overall responsiveness. If you do not have multiple systems to distribute load or if your systems are already load-balanced, skip to the sections of this chapter on tuning.
The following sections discuss techniques for load balancing each of the major system resources.
To load-balance memory you should try to ensure that page faulting and memory use are distributed proportionally among available systems. Monitor memory use on each system (during peak times) using the...